Evaluating acoustic deterrents for reducing migratory bird collisions with buildings

Document Type

Article

Publication Date

7-1-2026

Abstract

Migratory songbird populations are declining across North America, and collisions with glass-fronted buildings have become one of their leading causes of mortality. Finding practical ways to curb these collisions is a key conservation priority, but large-scale solutions remain elusive because retrofitting glass façades and enforcing lights-out policies are often prohibitively costly and logistically difficult. In this study, we focused on developing low-cost alternative deterrents that redirect migrants away from high-risk areas. We tested two acoustic deterrents: a broadband synthetic signal designed to mask nocturnal flight calls, which are important social cues for migrants, and a “predator” track featuring alarm calls and local raptor vocalizations. Their effects were first assessed at five forest-meadow stopover sites in Michigan’s Upper Peninsula during spring 2023. We used autonomous recorders recording migrant flight calls as a measure of migrant activity before and after deterrent playback. Autonomous recorders showed a 42% drop in migrant detections during synthetic playback and a 20% decline during predator playback. We then applied a before– after control–impact design on Northwestern University’s Lake Michigan campus, focusing on four buildings that historically register the most strikes. A hierarchical zero-inflated Poisson model controlling for migration intensity and building identity revealed that predator and synthetic broadcasts cut collisions by 56% and 68%, respectively, in 2023, when both units suffered occasional malfunctions. In 2024, predator deterrents operated continuously resulting in an 89% decrease in collisions. Our results suggest that acoustic deterrents represent a practical and scalable pathway to large reductions in passage migrant building collisions.

Publication Title

Avian Conservation and Ecology

Link to Full Text

Share

COinS